EP3422123B1 - Method and assembly for the computer-assisted configuring of a technical system - Google Patents

Description

For the operation of technical systems, for example power grids, wind turbines, gas turbines or manufacturing plants, it is often required that given technical parameters meet certain boundary conditions with a given probability under representative operating conditions. In this context, it is very useful for a configuration of a technical system to determine that value of a given parameter which, under representative operating conditions, will not be exceeded with a given probability. Accordingly, for a given point in a power grid, for example, that voltage value can be determined which, under representative operating conditions, has a probability of not more than one percent. Such a stress value is also referred to as a 1% quantile. The power grid can then be configured in such a way that the voltage value determined in each case satisfies the required boundary conditions with the required probability under representatively varying operating conditions.

In general, a q-quantile is to be understood as that parameter value in which a portion of q of the parameter values is below the q-quantile and the rest of the parameter values are above the q-quantile. The proportional value q of a q-quantile is often referred to as the undershoot portion. A 50% quantile is also known as the median.

Insofar as the operating conditions of a technical system are often subject to considerable, in particular random fluctuations, such as the wind conditions in a wind turbine, the operating parameters specifying the operating conditions are often modeled as random variables. A quantile of one that is dependent on statistically distributed operating parameters technical parameters can then be determined using a Monte Carlo simulation. A large number of simulations with randomly distributed operating parameters are carried out, an estimated value for the resulting technical parameter being determined in each case. The sought quantile of the technical parameter can then be determined from the large number of resulting estimated values.

In the case of sufficiently complex technical systems, however, an exact Monte Carlo simulation often proves to be very complex or hardly feasible due to the large number of simulation runs required. On the other hand, a less accurate simulation with a simplified simulation model is not accurate enough for many applications.

From the document " Common Rank Approximation- A new method to speed up probabilistic calculations in distrbtion grid planning "by M. Lindner and R. Witzmann for the" Transmission and Distribution Conference and Exposition (T&D), 2016 IEEE / PES "It is known to carry out a large number of simulations on the basis of a simplified simulation model and to use a more precise simulation model to improve the results. Although this method can considerably reduce the computational effort required, the method only provides an approximation of the quantile sought.

The object of the present invention is to specify a method, an arrangement, a computer program product and a computer-readable storage medium for the computer-aided configuration of a technical system, which allow a more precise configuration of the technical system.

This object is achieved by a method with the features of patent claim 1, by a method with the features of patent claim 10, by an arrangement with the features of patent claim 11, by a computer program product with the features of patent claim 12 and a computer-readable storage medium with the features of claim 13.

According to the invention, for the computer-aided configuration of a technical system, a proportional value specifying the availability of a provision parameter of the technical system is read in. In this case, the provision parameter can in particular indicate a property of a supply item made available in the technical system. The proportion value can in particular specify an underflow proportion of a quantile to be determined of the provision parameter. Furthermore, a large number of value variants of an operating parameter of the technical system is generated. For a respective value variant, a respective first estimated value for the provision parameter is determined by a first simulator of the technical system and assigned to the respective value variant. Furthermore, the value variants are at least partially sorted according to the respectively assigned first estimated values, and a quantile of the sorted value variants corresponding to the proportional value is determined. In addition, several value variants located in the vicinity of the quantile with regard to the sorting are selected. For a respective selected value variant, a respective second estimated value for the provision parameter is determined by a second simulator of the technical system. Alternatively or additionally, a respective measured value for the provision parameter is determined for a respective selected value variant. According to the invention, by compensating for fluctuations in the second estimated values or the measured values, a third estimated value for the provision parameter is determined and output as a configuration parameter for configuring the technical system.

To carry out the method according to the invention, an arrangement for the computer-aided configuration of the technical system, a computer program product and a computer-readable storage medium are provided.

The method according to the invention and the arrangement according to the invention can be executed or implemented, for example, with the aid of one or more processors, application-specific integrated circuits (ASIC), digital signal processors (DSP) and / or so-called "Field Programmable Gate Arrays" (FPGA).

One advantage of the invention is that the third estimated value can be used to determine a relatively precise value for the q-quantile of the provision parameter with relatively little computing effort. The technical system can thus be configured more precisely by means of the third estimated value and adapted to specified availability requirements.

The invention can in particular be used for setting up, adjusting, planning, designing, laying out, constructing and / or commissioning technical systems or their components.

Advantageous embodiments and developments of the invention are specified in the dependent claims.

According to an advantageous embodiment of the invention, a supply network with a plurality of network nodes can be configured as the technical system. In this case, the operating parameter can indicate an infeed, withdrawal or demand for a supply item at a network node, and the provision parameter a property or quality of the supply item made available at a network node or at a node connection. In particular, energy, gas, water or district heating can be provided as supply goods. The supply network can then accordingly be an energy network, gas distribution network, water distribution network or district heating network with power lines, gas lines, water lines or district heating lines as node connections. A voltage, a pressure, a level, a temperature or some other quantity that indicates how well a need is met can be specified.

In particular, a power network can be configured as the supply network. The operating parameter can indicate a feed-in power or extraction power at a network node and the provision parameter a voltage, an active power, a reactive power, an apparent power or a current at a network node or at a node connection.

The first simulator can preferably determine the first estimated values on the basis of a simulation model that is simplified compared to a simulation model of the second simulator. Because of this simplification, the simulation model of the first simulator can generally be evaluated with considerably fewer computing resources than the simulation model of the second simulator. In this way, the calculation of the first estimated values for the large number of value variants can generally be shortened considerably. The second estimated values can generally be calculated more precisely than by the first simulator using the simulation model of the second simulator, which is more complex to evaluate. However, insofar as the calculation of the second estimated values can be restricted to the selected value variants, the computational effort required for this generally remains acceptable. In this way, a relatively high level of accuracy can be combined with a relatively low computational effort.

A linearized simulation model can advantageously be used as the simplified simulation model. In particular, the simplified simulation model can be derived by linearizing a non-linear simulation model of the second simulator. Such linearized simulation models can usually be evaluated particularly efficiently, quickly and stably. A large number of standard programs are available for this.

According to an advantageous embodiment of the invention, the value variants can be generated in accordance with a predetermined frequency distribution and / or in accordance with a predetermined boundary condition. The frequency distribution can preferably indicate a statistical distribution of the values of the operating parameters to be expected under typical operating conditions of the technical system. In this way, the simulations can be used to simulate realistic and / or representative operating conditions.

According to an advantageous development of the invention, a fluctuation range of the second estimated values can be determined. On the basis of the fluctuation range, a measure for the accuracy of the third estimated value can then be derived and output for configuring the technical system. The fluctuation range can e.g. a statistical variance, scatter or standard deviation of the second estimated values can be determined. In this way, the accuracy of the third estimated value can be estimated as the quantile of the provision parameter that corresponds to the proportional value without significant additional effort.

According to an advantageous embodiment of the invention, the fluctuations can be compensated for using a compensation curve for the second estimated values. In particular, the third estimated value can be determined by evaluating the regression curve on the quantile of the sorted value variants. As a compensation curve, e.g. a straight line or a parabola can be used, the curve parameters of which can be determined with little effort using standard methods.

Figur 1

ein Stromnetz mit mehreren Netzknoten,

Figur 2

eine erfindungsgemäße Konfigurationseinrichtung zum Konfigurieren des Stromnetzes und

Figur 3

ein Diagramm zur Veranschaulichung einer Genauigkeit des erfindungsgemäßen Verfahrens

An exemplary embodiment of the invention is explained in more detail below with reference to the drawing. In each case show in a schematic representation:

Figure 1

a power grid with several network nodes,

Figure 2

a configuration device according to the invention for configuring the power grid and

Figure 3

a diagram to illustrate an accuracy of the method according to the invention

Figure 1 shows a power network SN as a technical system to be configured in a schematic representation. Such a power grid can include a variety of grid components, such as power generators, conventional or renewable energy sources, power plants, photovoltaic systems, wind turbines, consumer loads and power lines.

In the present exemplary embodiment, the power network SN has a conventional power plant G and wind power plants W as network nodes, which each function as a feed point. As a further network node, the power network SN includes a consumer load L as a withdrawal point. The network nodes G, W and L are connected by node connections C, here in particular power lines. A control device CTL, for example a so-called CPS (Control and Protection System), is provided to control the power network SN. The control device CTL is - as in Figure 1 indicated by dashed lines - coupled with the network nodes G, W and L.

Instead of the power network SN, other supply networks for supplies such as gas, water or district heating can also be provided. Here, too, network nodes, in particular feed and withdrawal points, are connected by node connections, in particular connecting or transport lines.

A complex technical system such as the power grid SN is characterized by a large number of operating parameters. In general, physical, control-related, effect-related and / or design-related operating parameters, properties, performance data, effect data, status data, system data, default values, control data, sensor data, measured values, environmental data, monitoring data, Forecast data, analysis data and / or other data occurring during the operation of the technical system and / or describing an operating state of the technical system are recorded.

In Figure 1 Power flows PG, PW and PL at network nodes G, W and L are shown as physical operating parameters. PG denotes a feed power from the power plant G, PW feed power from the wind power plants W and PL denotes a withdrawal power at the consumer network node L. The physical operating parameters PG, PW and PL are preferably represented as components of a vector P = (PG, PW, PL).

Depending on the physical operating parameters P, the power network SN provides the electrical power PL with a voltage U at the consumer network node L. In this context, the voltage U can be understood as a physical supply parameter which indicates a quality of the supplied goods, here the withdrawal rate PL.

The voltage U, with which the power PL is made available, is generally only allowed to lie outside of narrow, predetermined limits with a very low probability. In general, however, both the feed-in power PW from the wind power plants W and the withdrawal power PL at the load network node L are sometimes subject to considerable fluctuations. Despite a partial compensation by the conventional power station G, these fluctuations usually induce certain fluctuations in the voltage U.

The power network SN is now to be configured by means of the invention in such a way that the fluctuations in the voltage U remain within the permitted range as far as possible. For this purpose, a configuration device CONV is provided for the computer-protected configuration of the power network SN, in particular the network nodes G, W, L and / or the control device CTL. The configuration facility CONV is coupled to the power network SN, to the network nodes G, W, L and / or to the control device CTL. The configuration device CONV can be implemented as part of the power network SN or completely or partially external to the power network SN.

The configuration device CONV is intended to configure the power network SN such that the voltage U falls below a specific value U _Q with representatively fluctuating operating parameters P only with a predetermined, low probability of 0.1%, for example. For this purpose, for a first configuration of the power grid SN, the configuration device CONV determines that voltage value U _Q which, with a probability of 0.1%, is undershot. The 0.1% quantile of the voltage U is determined with U _Q. If the determined voltage value U _Q does not meet the required boundary conditions, the first configuration is changed and the associated voltage value U _{Q is} determined again until the required boundary conditions are met.

Figure 2 shows a configuration device CONV according to the invention for configuring the power network SN in a schematic representation. The configuration device CONV comprises one or more processors PROC for carrying out all work steps of the configuration device CONV and one or more memories MEM for storing data to be processed by the configuration device CONV. The configuration device CONV is coupled to the control device CTL of the technical system SN.

As already mentioned above, in the present exemplary embodiment the infeed powers PG and PW and the withdrawal power PL are considered as physical operating parameters of the power grid SN and are combined in the vector P. Because of their often strong random fluctuations, these operating parameters are advantageously regarded as random variables.

The voltage U at the consumer network node L is considered as a physical provision parameter. Alternatively or additionally, an active power, a reactive power, an apparent power and / or a current at one or more network nodes or node connections can also be considered as provision parameters of the power network SN. Several provision parameters are represented as components of a vector if necessary. The provision parameter (s), here U, each depend deterministically on the operating parameters P via a network dynamics of the power network SN. The fluctuations in the operating parameters P therefore generally induce corresponding fluctuations in the provision parameters, here U.

The configuration device CONV reads in a statistical frequency distribution DST of the operating parameters P considered as random variables and a proportional value q. The proportion value q here specifies an availability of the provision parameter U, insofar as the proportion value q specifies or limits that proportion of values of the provision parameter U which is below the q-quantile to be determined.

In the present exemplary embodiment, a proportional value of q = 0.1% is read in, as a result of which the 0.1% quantile U _{Q of} the voltage U is determined. The proportional value q is transmitted to a selection module SEL of the configuration device CONV.

The frequency distribution DST is intended to realistically and representatively indicate an expected statistical distribution of the values of the operating parameters P. The frequency distribution DST can be measured empirically in the technical system SN or determined in some other way. Preferably, one or more boundary conditions that are to be met by the operating parameters P can also be read in with the frequency distribution DST.

The frequency distribution DST is transmitted together with the boundary conditions to a variant generator GEN of the configuration device CONV. The variant generator GEN is used for the random-based generation of a large number of possible value variants P _{I of} the operating parameters P. The value variants P _I are generated depending on the frequency distribution DST that has been read in and, if applicable, depending on the boundary conditions that have been read in such that they themselves have the frequency distribution DST and, if applicable, the read in boundary conditions are sufficient. A large number of efficient standard methods in the context of so-called Monte Carlo simulations are available for generating such value variants.

The generated value variants P _I are indexed by the index I. In the present exemplary embodiment, a number of 100,000 value variants P _{I are} generated, ie I = 1,..., 100,000. The value variants P _I are preferably implemented by an array of operating parameters or operating parameter vectors, which is indexed by the index I.

In a number of 100000 variants value P _I is the one to be detected 0.1% quantile U _Q voltage value of the induced by the operating parameters P _I variants voltages U (P _I) below, and the remainder are in the 100 above. The value variants P _I are transmitted from the variant generator GEN to a first simulator SIM1 of the configuration device CONV, which is coupled to it.

In addition to the first simulator SIM1, the configuration device CONV has a second simulator SIM2. The simulators SIM1 and SIM2 each simulate a behavior of the technical system SN induced thereby for different value variants P _I. Within the scope of the simulations, the first simulator SIM1 determines, for a respective value variant P _I, in particular a respective first estimated value U1 (P _I ) for the provision parameter U induced by the respective value variant P _I. The second simulator determines accordingly SIM2 for selected value variants P _{I in} each case a second estimated value U2 (P _I ) for the provision parameter U induced by the relevant value variant P _I.

The first simulator SIM1 has a first simulation model SM1 of the technical system SN and the second simulator SIM2 has a second simulation model SM2 of the technical system SN. The simulations of the first simulator SIM1 and of the second simulator SIM2 are each carried out using the first simulation model SM1 and the second simulation model SIM2.

The second simulation model SM2 is preferably a detailed, generally non-linear simulation model of the technical system SN. In the present exemplary embodiment, the second simulation model SM2 is a non-linear load flow model of the power network SN, which is based on a system of non-linear load flow equations of the power network SN. A simulation based on such a non-linear simulation model generally requires a relatively high computing effort. In the present case, an individual simulation using the second simulator SIM2 for all 100,000 value variants P _I would hardly be acceptable with regard to the computational effort required for many applications.

The first simulation model SM1 of the power grid SN is a simplified or reduced simulation model compared to the second simulation model SM2. Such a simplification or reduction can preferably be achieved by linearizing the load flow equations on which the second simulation model SM2 is based. A simulation based on a linearized simulation model, here SM1, generally requires considerably fewer computing resources than a simulation based on a nonlinear simulation model, here SM2. In particular, using the first simulation model SM1, the first estimated values U1 (P _I ) can usually be determined for all value variants P _I , that is to say here for 100,000 value variants, with acceptable effort.

The first simulation model SM1 can function as a substitute model for the second simulation model SM2 in many cases. However, a simulation using the first simulation model SM1 is generally less accurate than a simulation using the second simulation model SM2.

The load flow equations for the first simulation model SM1 can be linearized, for example, by neglecting higher terms of a deviation of a network voltage from a reference network voltage, setpoint network voltage or another predetermined network voltage.

When creating the first simulation model SM1 or when deriving the first simulation model SM1 from the second simulation model SM2, the aim is in particular that when sorting the second estimated values U2 determined using the second simulation model SM2 according to their size, the same sequence should ideally result as when sorting the estimated values U1 determined by means of the first simulation model SM1 according to their size. Although the above sorting condition cannot usually be met exactly, the sortings of U1 and U2 often show at least a certain tendency to match. It has been found that this tendency of agreement is already sufficient in many cases to considerably reduce the number of detailed simulations required by the second simulator SIM2.

As already mentioned above, the first estimated values U1 (P _I ) are determined by the first simulator SIM1 on the basis of the first simulation model SM1 by simulating a dynamic of the power grid SN. The first estimated values U1 (P _I ) determined are then transmitted together with the value variants P _I by the first simulator SIM1 to a sorting module SORT of the configuration device CONV that is coupled to this.

Using the first estimated values U1 (P _I ), the sorting module SORT sorts the array of the value variants P _I according to the size of the first estimated values U1 (P _I ). This rearrangement of the array P _I results in an array of sorted value variants PS _K with a sorting index K. The sorting index K = 1, ..., 100000 indexes the value variants PS _K sorted according to the first estimated values in such a way that the correspondingly sorted first estimated values U1 (PS _K ) increase monotonically with increasing sorting index K.

The sorted first estimated values U1 (PS _K ) are transmitted together with the sorted value variants PS _K from the sorting module SORT to a selection module SEL of the configuration device CONV that is coupled to this. The selection module SEL is used to select a q-quantile QP of the sorted value variants PS _K and to select an area IVL of the q-quantile QP. The proportional value q, here 0.1%, is read in by the SEL selection module.

The q-quantile QP with regard to the sorting of the sorted value variants PS _{K is} determined as a function of the read-in share value q, ie in such a way that a share of q of the sorting indices K is below the quantile index, the rest above. With 100,000 sorted value variants PS _K and a proportional value of q = 0.1%, QP = PS ₁₀₀ .

The preferably symmetrical environment IVL of the quantile QP is also selected with regard to the sorting of the sorted value variants PS _K. According to the invention, a number of value variants is selected from the sorted value variants PS _K which is small compared to the number of all value variants P _I. In the present exemplary embodiment, an environment IVL with 101 sorted value variants PS _{K is} selected, so that IVL = (PS ₅₀ , ..., PS ₁₅₀ ). In contrast, the number of all value variants is several orders of magnitude larger.

The number of selected value variants PS ₅₀ , ..., PS ₁₅₀ determines the number of required, complex simulations through the second simulator SIM2. In this way, in the present exemplary embodiment, the number of complex simulations by the second simulator SIM2 can be reduced to 1/1000 of the number of simulations by the simplified, first simulator SIM1.

The q-quantile QP and its environment IVL are transmitted from the selection module SEL to the second simulator SIM2 coupled to it. The second simulator SIM2 determines second estimated values U2 (PS _K ) for K = 50, ..., 150 for the selected value variants PS ₅₀ , ..., PS _{150 of} the environment IVL. No complex simulation is carried out by the second simulator SIM2 for value variants outside the environment IVL.

The second estimated values U2 (PS _K ) are preferably stored as an array sorted according to K or U1 (PS _K ). In Figure 2 the second estimated values (U2 (PS ₅₀ ), ..., U2 (PS ₁₅₀ )) are designated as U2 (IVL).

The second estimated values U2 (PS _K ) sorted according to the sorting index K can show considerable fluctuations as a function of K, since they are sort of sorted according to the size of the first estimated values U1 (PS _K ), and the sorting of the first and second estimated values - as mentioned above - usually do not match exactly.

The q-quantile QP and the second estimated values U2 (IVL) are transmitted from the second simulator SIM2 to a compensation module AM of the configuration device CONV, which is coupled to this. The compensation module AM is used to compensate for fluctuations in the second estimated values U2 (PS _K ) sorted according to the first estimated values. For this purpose, the second estimated values U2 (PS _K ) can be statistically averaged. Alternatively or additionally, a model or a model curve can be adapted to a course of the second estimated values U2 (PS _K ) in such a way that a deviation between the model or model curve and the second estimated values U2 (PS _K ) is as small as possible. For such an adjustment that often Also known as fitting or regression, a variety of standard methods are available.

In the present exemplary embodiment, the fluctuations in the second estimated values U2 (PS _K ) are compensated for by a compensation curve AK, for example a compensation parabola. The curve parameters of the compensation curve AK are determined in such a way that deviations between the compensation curve AK and the second estimated values U2 (PS _K ) for K = 50,..., 150 are minimal. The compensation curve AK is evaluated by the compensation module AM at the quantile QP of the selected value variants, that is, when K = 100. The result of this evaluation is a third estimated value U _Q = AK (PS ₁₀₀ ). The third estimated value U _Q is thus a value for the q-quantile sought for the provision parameter U.

It has been found that the majority of the errors that have arisen as a result of the above-mentioned sorting condition not being met can be compensated for by the compensation curve AK. The q-quantile U _Q can therefore usually be determined very precisely by means of the fluctuation compensation according to the invention.

Furthermore, the compensation module AM determines a fluctuation range of the second estimated values U2 (IVL), for example as the mean deviation between the compensation curve AK and the curve of the second estimated values U2 (IVL). As an alternative or in addition, a variance, scatter or standard deviation of the second estimated values U2 (IVL) can also be determined as the fluctuation range. The fluctuation range forms a measure DU for an accuracy of the third estimated value U _Q , which can be determined within the scope of the method according to the invention without great additional effort.

The third estimated value U _Q and the degree of accuracy DU are transmitted as configuration parameters from the configuration device CONV to the control device CTL for configuring the power network SN.

Figure 3 shows a diagram to illustrate an accuracy of the method according to the invention. The diagram shows a course of the quantities described above, namely the sorted first estimated values U1 (PS _K ), the second estimated values U2 (PS _K ) and the compensation curve AK, each as a function of the sorting index K for K = 50, ..., 150 shown. A compensation parabola was chosen as the best fit curve AK.

The sorting index K is plotted on the abscissa of the diagram and the voltage U at the consumer network node L in a so-called per-unit system is plotted on the ordinate. Such a per-unit system, abbreviated “pu”, specifies the voltage U in relation to a predetermined reference value.

Since the sorted value variants PS _{K are} sorted according to the size of the first estimated values U1, U1 (PS _K ) is a function of K - as in FIG Figure 3 can be seen - monotonically increasing. In contrast, the second estimated values U2 (PS _K ) as a function of K - as already explained above - exhibit considerable fluctuations. In addition, it can be clearly seen that the second estimated values U2 (PS _K ) determined by the more precise second simulator SIM2, regardless of their fluctuations, systematically deviate from the first estimated values U1 (PS _K ) determined by the first simulator SIM1.

The fluctuations in the second estimated values U2 (PS _K ) are largely compensated for by the compensation parabola AK. The third estimated value U _Q is obtained by evaluating the compensation parabola AK at the q-quantile QP.

To illustrate the accuracy of the third estimated value U _Q is in Figure 3 a course of second estimated values US2 determined by the second simulator SIM2 is also shown. The second estimated values US2 are shown sorted according to their own size. The sorting of the second estimated values US2 results in a monotonic course.

An evaluation of the course of the second estimated values US2 on the proportional value q results in a q-quantile QUS2, here a 0.1% quantile, of the second estimated values US2. Insofar as the second estimated values US2 were determined by the more precise second simulator SIM2, the q-quantile QUS2 can be understood as a relatively precise reference value for the q-quantile of the voltage U sought.

It should be noted that the representation of the second estimated values US2 in Figure 3 serves only for comparison purposes, insofar as a complex simulation is to be carried out by the second simulator SIM2 for all 100,000 value variants to determine them, while according to the invention the complex simulation can be limited to the considerably smaller area of the environment IVL. As already mentioned above, in the present exemplary embodiment the number of complex simulations by the second simulator SIM2 can be reduced to 1/1000 of the number of simplified simulations by the first simulator SIM1.

It can be seen that the third estimated value U _Q deviates only very slightly from the q-quantile QUS2 determined by precise simulation. U _Q is therefore a very good estimated value that can be determined with considerably less effort than the q-quantile QUS2. By means of the invention, a high degree of accuracy can be achieved with a relatively low computational effort. In addition, the variable DU also provides a measure of the accuracy of the determined third estimated value U _Q and can advantageously be taken into account in the configuration of the technical system SN.

Claims (13)

1. Method for the computer-aided configuration of a technical system (SN), wherein

   a) a proportion value (q) specifying an availability of a provision parameter (U) of the technical system (SN) is read in,

   b) a multiplicity of value variants (P_I) of an operating parameter (PG, PW, PL) of the technical system (SN) are generated,

   c) a respective first estimate (U1) for the provision parameter (U) is determined for a respective value variant (P_I) by a first simulator (SIM1) of the technical system (SN) and is assigned to the respective value variant (P_I),

   d) at least some of the value variants (P_I) are sorted according to the respectively assigned first estimates (U1),

   e) a quantile (QP) for the sorted value variants (PS_K) that corresponds to the proportion value (q) is determined,

   f) multiple value variants (PS₅₀,...,PS₁₅₀) that are in a vicinity (IVL) of the quantile (QP) in respect of the sorting are selected,

   g) a respective second estimate (U2) for the provision parameter (U) is determined for a respective selected value variant (PS₅₀,...,PS₁₅₀) by a second simulator (SIM2) of the technical system (SN),

   h) a third estimate (U_Q) for the provision parameter (U) is determined by equalizing variations in the second estimates (U2), and

   i) the third estimate (U_Q) is output as a configuration parameter for configuring the technical system (SN).
2. Method according to Claim 1, characterized
   in that the technical system (SN) configured is a supply network having multiple network nodes,
   in that the operating parameter specifies a feed of, an extraction of or a need for a supply item at a network node, and
   in that the provision parameter specifies a property or quality level of the supply item provided at a network node or at a node connection.
3. Method according to Claim 2, characterized
   in that the supply network configured is a power grid (SN),
   in that the operating parameter specifies a feed power (PG, PW) or extraction power (PL) at a network node (G, W, L), and in that the provision parameter specifies a voltage (U), a real power, a reactive power, an apparent power or a current at a network node (G, W, L) or at a node connection (C).
4. Method according to one of the preceding claims, characterized in that
   the first simulator (SIM1) determines the first estimates (U1) on the basis of a simulation model (SM1) that is simplified in comparison with a simulation model (SM2) of the second simulator (SIM2).
5. Method according to Claim 4, characterized in that
   the simplified simulation model (SM1) used is a linearized simulation model.
6. Method according to one of the preceding claims, characterized in that
   the value variants (P_I) are generated in accordance with a prescribed frequency distribution (DST) and/or in accordance with a prescribed constraint.
7. Method according to one of the preceding claims, characterized
   in that a variability of the second estimates (U2) is determined and
   in that the variability is taken as a basis for deriving a measure (DU) of an accuracy of the third estimate (U_Q) and outputting it for configuring the technical system (SN).
8. Method according to one of the preceding claims, characterized
   in that the variations are equalized on the basis of an equalization curve (AK) for the second estimates (U2).
9. Method according to Claim 8, characterized
   in that the third estimate (U_Q) is determined by evaluating the equalization curve (AK) using the quantile (QP) of the sorted value variants (PS_K).
10. Method for the computer-aided configuration of a technical system (SN), wherein

    a) a proportion value (q) specifying an availability of a provision parameter (U) of the technical system (SN) is read in,

    b) a multiplicity of value variants (P_I) of an operating parameter (PG, PW, PL) of the technical system (SN) are generated,

    c) a respective first estimate (U1) for the provision parameter (U) is determined for a respective value variant (P_I) by a first simulator (SIM1) of the technical system (SN) and is assigned to the respective value variant (P_I),

    d) at least some of the value variants (P_I) are sorted according to the respectively assigned first estimates (U1),

    e) a quantile (QP) for the sorted value variants (PS_K) that corresponds to the proportion value (q) is determined,

    f) multiple value variants (PS₅₀,...,PS₁₅₀) that are in a vicinity (IVL) of the quantile (QP) in respect of the sorting are selected,

    g) a respective measured value for the provision parameter (U) of the technical system (SN) is determined for a respective selected value variant (PS₅₀,...,PS₁₅₀),

    h) a third estimate (U_Q) for the provision parameter (U) is determined by equalizing variations in the measured values, and

    i) the third estimate (U_Q) is output as a configuration parameter for configuring the technical system (SN).
11. Arrangement for the computer-aided configuration of a technical system, designed to carry out a method according to one of the preceding claims.
12. Computer program product designed to carry out a method according to one of Claims 1 to 10.
13. Computer-readable storage medium having a computer program product according to Claim 12.

Images